Satellite-based Spatio-temporal Variation Of Cabon Storage Of Moso Bamboo Forest In Anji Country

Bamboo forest is a special forest type mainly distributed in semi-tropical areas of China. It has been widely confirmed that bamboo forests with huge biomass or carbon storage play important roles in protecting regional ecological environment and maintaining global carbon balance. As studies on bamboo forests increased, many methods used to extract bamboo forest information based on remote sensing data were proposed and a number of publications focus on estimating bamboo biomass and its spatial distribution. However, study on the spatial autocorrelation and heterogeneous of Moso bamboo aboveground biomass (AGB)/ aboveground carbon (AGC) storage in a large scale and time-series is rare. Therefore, it is important to evaluate and analyze the spatial change characteristic of bamboo forest AGB/AGC, and this study is significant.In this study, based on remote sensing images with five different times, RS, GIS and GS+ were used to study spatial heterogeneous characteristics and changes in carbon storage of Moso bamboo in Anji, Zhejiang province, China. This study includes the following aspects:1. Maximum likelihood method was used to extract Moso bamboo forest information based on Landsat TM images.2. Rangeability and dynamical degree were used to analyze the spatial dynamic change character of bamboo forest. Effects of land cover change on bamboo forest were been analyzed.3. Based on the integration of Landsat TM and field inventory data, aboveground carbon storage estimation model was built and then used to estimate aboveground carbon storage in different periods.4. The semivariogram of carbon storage were calculated using geostatistical software GS+ and these graphs were fitted using different statistical models. The optimal model was used to analyze the spatial heterogeneity of Moso bamboo forest carbon storage in five periods.This study mainly gets the following conclusions:1. Classification accuracy of each image and Moso bamboo forest was high enough. The overall classification accuracy of each TM image was higher than 85%, and kappa coefficient of Moso bamboo forest ranged from 0.80 to 0.95. The classification accuracy is suitable for application.2. Except decrease in bamboo forest area of Kuntong (rangeability is -8.49%), the bamboo forest areas of other towns were increasing during 1986 and 2008, and the rangeability ranged from 14% to 86%. The rangeability of Xiaofeng was the largest, and the rangeability of Tianhuangping was the smallest.3. Area percent of Moso bamboo forests in Anji increased from 28.89% in 1986 to 38.48% in 2008. The main reason is that the conversions of conifer and broad-leave forests to bamboo forests.4. Results showed that carbon storage of Moso bamboo forests notablely increased from 1986 to 2008. The aboveground carbon density in 5 different periods were 11.50Mg/ha, 15.09 Mg/ha, 18.35 Mg/ha, 23.41 Mg/ha, 22.46 Mg/ha, respectively. Except that spatial distribution of carbon density in 1986 is even, there are similarities in carbon density spatial distribution for the other four periods. Large carbon density exists in southwest and southeast and small carbon density exists in northeast and northwest. Increase rate of carbon density in each town is over 50% from 1986 to 2008.5. Results of semi-variance function simulation showed that the optimal semi-variance function for 1991 is spherical model and the optimal semi-variance functions for the other four periods are exponential models.6. Analysis on theoretical model structure showed that a range of spatial autocorrelation in 1986 was the largest (9870 m) and it obviously decreased ranged between 2000 and 3000m after 1991. Spatial autocorrelation characters of carbon density in five periods were medium, and their spatial structure ratio (C/C0+C) were 50.1%,72.4%,66.8%,67.3% and 65.6% respectively. It implies that those structural factors, such as terrain, soil, climate, play an important role in spatial heterogeneity of Moso bamboo forest carbon storage. Except for 1986, ratio of random factors increased from 1991 to 2008, that is, ratio of random variance (C0) increased. The result is consistent with Moso bamboo forests severely disturbed by human management in recent years.